Enterprise cloud garbage collector
Various embodiments collect unproductive resources in a network infrastructure. In one embodiment, data relating to resources of a network infrastructure is collected. An analytics model is selected based on a type of the collected data. The selected analytics model is executed to classify a resource unproductive or productive, and to assign a corresponding confidence level. An action plan for each confidence level is determined and the action plan is executed for the resource. The collected data may include resource utilization information, hypervisor information, cloud related meta-data, user knowledge and system knowledge. When data is only resource data, a resource mining model is selected. When the data includes reference data, a reference mining model is selected. When the data comprises reference data and resource data, a combined mining model is selected.
1. A method for collecting unproductive resources in a network infrastructure, the method comprising:
collecting data relating to resources of a network infrastructure;
selecting an analytics model based on a type of the collected data;
executing the selected analytics model to classify a resource as one of unproductive and productive and to assign a corresponding confidence level;
determining an action plan for each confidence level; and
executing the action plan for the resource.
2. The method of claim 1 , wherein the collected data comprises at least one of resource utilization information, hypervisor information, cloud related meta-data, user knowledge and system knowledge.
3. The method of claim 1 , wherein the data type comprises only resource data, the selected analytics model is a resource mining model.
4. The method of claim 1 , wherein the data type comprises reference data, the selected analytics model is a reference mining model.
5. The method of claim 4 , wherein the reference mining model:
generates a reference graph of resources within the network infrastructure;
assigns a weight to each edge of the reference graph based on a set of heuristic rules; and
determines the confidence level for each resource based on reachability.
6. The method of claim 1 , wherein the data type comprises reference data and resource data, the selected analytics model is a combined mining model.
7. The method of claim 6 , wherein the combined mining model:
generates a reference graph of resources within the network infrastructure;
assigns a weight to each edge of the reference graph based on a set of heuristic rules;
classifies each resource as one of unproductive and productive with a corresponding confidence level for each resource based on reachability;
reclassifies each resource as one of unproductive and productive with a corresponding probability level based on resource utilization; and
modifies the confidence level based on the reclassification and corresponding probability level.
8. The method of claim 7 , wherein a resource is originally classified as productive, the method further comprises ignoring the reclassification.
9. The method of claim 7 , wherein a resource is originally classified as unproductive the method further comprises downgrading the confidence level when the resource is reclassified as productive.
10. The method of claim 7 , wherein a resource is originally classified as unproductive the method further comprises upgrading the confidence level when the reclassifying affirms that the resource is classified as unproductive.
11. The method of claim 1 , wherein the action plan includes at least one of: notifying a user that a resource is unproductive, suspending the resource, decommissioning the resource and reallocating the resource.
12. An information processing system for collecting unproductive resources in a network infrastructure, the information processing system comprising
a memory;
a processor operably coupled to the memory; and
an enterprise garbage collector operably coupled to the memory and the processor, the enterprise garbage collector configured to perform a method comprising:
collecting data relating to resources of a network infrastructure;
selecting an analytics model based on a type of the collected data;
executing the selected analytics model to classify a resource as one of unproductive and productive and to assign a corresponding confidence level;
determining an action plan for each confidence level; and
executing the action plan for the resource.
13. The information processing system of claim 12 , wherein the collected data comprises at least one of resource utilization information, hypervisor information, cloud related meta-data, user knowledge and system knowledge.
14. The information processing system of claim 12 , wherein the data type comprises reference data, the selected analytics model is a reference mining model.
15. The information processing system of claim 14 , wherein the reference mining model:
generates a reference graph of resources within the network infrastructure;
assigns a weight to each edge of the reference graph based on a set of heuristic rules; and
determines the confidence level for each resource based on reachability.
16. The information processing system of claim 12 , wherein the data type comprises reference data and resource data, the selected analytics model is a combined mining model.
17. The information processing system of claim 16 , wherein the combined mining model:
generates a reference graph of resources within the network infrastructure;
assigns a weight to each edge of the reference graph based on a set of heuristic rules;
classifies each resource as one of unproductive and productive with a corresponding confidence level for each resource based on reachability;
reclassifies each resource as one of unproductive and productive with a corresponding probability level based on resource utilization; and
modifies the confidence level based on the reclassification and corresponding probability level.
18. The information processing system of claim 17 , wherein a resource is originally classified as productive, the method further comprises ignoring the reclassification.
19. The information processing system of claim 17 , wherein a resource is originally classified as unproductive the method further comprises downgrading the confidence level when the resource is reclassified as productive.
20. A computer program product for collecting unproductive resources in a network infrastructure, the computer program product comprising:
a storage medium readable by a processing circuit and storing instructions for execution by the processing circuit for performing a method comprising:
collecting data relating to resources of a network infrastructure;
selecting an analytics model based on a type of the collected data;
executing the selected analytics model to classify a resource as one of unproductive and productive and to assign a corresponding confidence level;
determining an action plan for each confidence level; and
executing the action plan for the resource.